def item_claim(nc=1, extra=0):
s = ''.join(random.choice(letters) for i in range(30))
if extra == 0:
q, r = [], []
if extra == 1:
q, r = [snak('string', s, 'P1')], []
if extra == 2:
q, r = [snak('string', s, 'P1')], [snak('string', s, 'P1')]
edict = entity(labels={'en': {
'language': 'en',
'value': s
}},
aliases={'en': [{
'language': 'en',
'value': s
}]},
descriptions={'en': {
'language': 'en',
'value': s
}},
claims=claim('P1',
snak('string', s, 'P1'),
qualifiers=q,
references=r),
etype="item",
datatype='string')
if nc >= 2:
for n in range(2, nc + 1):
edict['claims'].update(
claim('P' + str(n),
snak('string', s, 'P' + str(n)),
qualifiers=q,
references=r))
return edict
def property_wd_all():
""" Send a SPARQL-request to Wikidata, get all properties with labels,
aliases, descriptions and datatypes, and return a list of JSON
representations of the properties"""
props = get_wikidata_properties(language='en')
props = props.replace(np.nan,
'').replace("http://www.wikidata.org/entity/",
"",
regex=True)
props = props.groupby([
'propertyWikidata', 'propertyType', 'propertyLabel',
'propertyDescription'
]).agg(
propertyAlias=('propertyAlias', lambda x: list(x)),
fURL=('fURL', lambda x: list(x)),
cURI=('cURI', lambda x: list(x)),
).reset_index()
props = props[props.propertyWikidata != ("P1630" or "P1921")]
props = props.sort_values(
by=['propertyWikidata'],
key=lambda col: col.str.replace('P', '', regex=True).astype(float))
plist = []
for ind, row in props.iterrows():
url, dtype, title, desc, alia, furl, curi = row
p = entity(labels=label(value=title),
aliases=alias(value=alia),
descriptions=description(value=desc),
claims=claim(prop='P1',
mainsnak=snak(datatype='external-id',
value=url,
prop='P1',
snaktype='value')),
etype='property',
datatype=datatypes[dtype])
for f in furl:
if f != '':
p['claims'].update(
claim(prop='P2',
mainsnak=snak(datatype='string',
value=f,
prop='P2',
snaktype='value')))
for c in curi:
if c != '':
p['claims'].update(
claim(prop='P3',
mainsnak=snak(datatype='string',
value=c,
prop='P3',
snaktype='value')))
plist.append(p)
return plist
def property_hrn(prop=''):
"""Create a simplified JSON represetation (only the label, aliases,
description and one claim) of the first local property 'Native company number'"""
p = entity(
labels=label(value='Native company number'),
aliases=alias(value=[
"Court, code and number",
'Register, type of register and company number',
'Registergericht, Registerart and Handelsregister number '
]),
descriptions=description(
value=
"The registration authority (Amtsgericht), the registration code and the registration number"
),
claims={},
etype='property',
datatype='external-id')
return p
def property_wd(prop=''):
"""For the given PID of a property in Wikidata returns its
simplified JSON represetation (only the label, aliases, description
and one claim)"""
if prop:
r = requests.get('https://www.wikidata.org/entity/' + prop +
'.json').json()
po = r.get('entities').get(prop)
p = entity(labels=restrict(po.get('labels')),
aliases=restrict(po.get('aliases')),
descriptions=restrict(po.get('descriptions')),
claims=claim(prop='P1',
mainsnak=snak(datatype='external-id',
value=prop,
prop='P1',
snaktype='value')),
etype='property',
datatype='string')
else:
p = None
return p
def property_wid(prop=''):
"""Create a simplified JSON represetation (only the label, aliases,
description and one claim) of the first local property 'Wikidata ID'"""
p = entity(labels=label(value='Wikidata ID'),
aliases=alias(value=["WID", 'WikidataID']),
descriptions=description(value="ID of an entity in Wikidata"),
claims=claim(prop='P2',
mainsnak=snak(
datatype='string',
value='http://www.wikidata.org/entity/$1',
prop='P2',
snaktype='value')),
etype='property',
datatype='external-id')
p['claims'].update(
claim(prop='P3',
mainsnak=snak(datatype='string',
value='http://www.wikidata.org/entity/$1',
prop='P3',
snaktype='value')))
return p
native_company_number, registered_office, registrar,
register_art, register_nummer, former_registrar,
register_flag_, register_flag_Note,
registerNummerSuffix, register_flag_Status_information
] = line
e = entity(
labels=label(value=name),
aliases={},
descriptions={},
claims=claim(
prop=opencorporatesid,
mainsnak=snak(datatype='external-id',
value='de/' + company_number,
prop=opencorporatesid,
snaktype='value'),
qualifiers=[
snak("time", [
'+' + retrieved_at, 0, 11,
'http://www.wikidata.org/entity/Q1985727'
], retrieved, 'value')
],
references=[
snak('external-id', 'de/' + company_number,
opencorporatesid, 'value')
]),
etype='item')
if registered_address:
e['claims'].update(
claim(
prop=streetaddress,
mainsnak=snak(datatype='monolingualtext',